-
-
Notifications
You must be signed in to change notification settings - Fork 487
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
FriCAS spkg-configure and Feature #35838
Conversation
The doctests marked |
FriCAS people say using SBCL instead of ECL gives 3 to 4 times faster
system.
…On Tue, 27 Jun 2023, 21:45 François Bissey, ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In build/pkgs/fricas/distros/gentoo.txt
<#35838 (comment)>:
> @@ -0,0 +1 @@
+sci-mathematics/fricas[sbcl]
Why do you impose it to be build with sbcl. Sure it is probably the
fastest lisp around, but is there a specific reason that you need to have
it?
—
Reply to this email directly, view it on GitHub
<#35838 (review)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAJXYHF5OKNYACSAM6BVC6LXNNA47ANCNFSM6AAAAAAZUN3DRE>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
I am not saying to make it ECL, I am saying that lips user on Gentoo make their own choices unless there is a technical reason (as in a breaking reason) to do so. If they are interested in speed they will make that choice themselves. |
When you build the This is a Sage-the-distribution-specific mechanism that we want to get rid of altogether (see #35856), for the benefit of downstream packagers and also for the pip-installed setting. See also #35858 |
OK, I'm trying now the patch https://github.com/sagemath/sagetrac-mirror/commit/73ad696e35101a24c6c5b03adb518fad8ef3a91b.diff from #33575 (it applies with a fizz) |
OK, feature seems to be implemented now, please have a look |
@@ -0,0 +1,16 @@ | |||
SAGE_SPKG_CONFIGURE( | |||
[fricas], [ | |||
AC_CACHE_CHECK([for FriCAS >= 1.3.8], [ac_cv_path_FRICAS], [ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the indentation is a bit wild
Has using system fricas been tested on more than 1 machine? |
on 2 different machines, 3 different configurations (using ECL as well as using SBCL as lisp compiler) |
Build&Test fails with lots of errors originating from the feature thing |
works for me locally, it's probably unrelated.
unless these errors come from unavailable fricas.
…On Fri, 30 Jun 2023, 23:41 Matthias Köppe, ***@***.***> wrote:
Build&Test fails with lots of errors originating from the feature thing
—
Reply to this email directly, view it on GitHub
<#35838 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAJXYHDZ6U6SSB2HLWPUGITXN5I3NANCNFSM6AAAAAAZUN3DRE>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
Yes, in the test run on GH Actions there is no FriCAS because it's an optional package... |
well, indeed - I suppose it's due to #33575 pointing to an outdated design, compared to e.g. what's in Another potential problem is |
sage.interfaces.abc |
I'll take a look what needs to be fixed in the feature definition later today. |
I'm actually working on this now. I'll post an update in an hour or two |
38d728e
to
47073ef
Compare
here is a non-working update. I'm lost at what
|
The documentation is being added in this PR, no? |
47073ef
to
f90bbce
Compare
Do we need |
If FriCAS is not available (I just remove the executable), I get doctest errors like
no idea what's wrong. Could it be due to The branch seems to work (modulo a number of maths doctest errors) if FriCAS is available. |
5eaf9af
to
5adc1b7
Compare
This is needed as SBCL-built FriCAS does not output anything if )quit is passed via -eval option, as opposed to ECL-built FriCAS.
044f6c1
to
870fa22
Compare
Documentation preview for this PR (built with commit 870fa22; changes) is ready! 🎉 |
Usually not a good idea to do a rebase that includes the dependency PR. |
All good, it seems. I can't officially review this, so, please... |
@mantepse - this allows using system-wide FriCAS. |
As FriCAS is only used as an executable, this is straightforward; quite a number of systems has Fricas 1.3.8, so this is useful, too. The FriCAS pexpect interface now uses the new `Executable` feature to determine the absolute pathname of the fricas executable (unless executed remotely). This is made possible by a simple refactor of the `sage.interfaces.expect.Expect` class: Computing the effective command line is no longer done in `set_server_and_command` (called by `__init__`); it is delayed until an interface is started and needs the command line. - Fixes sagemath#35837 - Fixes sagemath#33575 Dependencies: The changes in `sage.interfaces` outside of `.expect` and `.fricas` are all from the following PR and do not need review. - Depends on sagemath#36656 (merged here) URL: sagemath#35838 Reported by: Dima Pasechnik Reviewer(s): Dima Pasechnik, François Bissey, Matthias Köppe
fricas 1.3.10 is meanwhile out. |
As FriCAS is only used as an executable, this is straightforward; quite a number of systems has Fricas 1.3.8, so this is useful, too.
The FriCAS pexpect interface now uses the new
Executable
feature to determine the absolute pathname of the fricas executable (unless executed remotely).This is made possible by a simple refactor of the
sage.interfaces.expect.Expect
class: Computing the effective command line is no longer done inset_server_and_command
(called by__init__
); it is delayed until an interface is started and needs the command line.Dependencies: The changes in
sage.interfaces
outside of.expect
and.fricas
are all from the following PR and do not need review.sage.interfaces
: Update# needs
#36656 (merged here)